Peptides for Chronic Fatigue Syndrome (ME/CFS): What the Evidence Actually Shows

Elamipretide is the only candidate with a human randomized controlled trial that directly tests the mitochondrial theory ME/CFS rests on, which is why it ranks first — and why it is the most instructive cautionary tale. In a randomized, double-blind, placebo-controlled trial in 39 healthy older adults (60–85 y) selected for poorly functioning mitochondria, a single two-hour IV infusion significantly raised in-vivo muscle mitochondrial ATP-production capacity (ΔATPmax p=0.045) — the first demonstration of a drug acutely reversing mitochondrial dysfunction in aging human muscle. But there was no significant effect on fatigue resistance, and the benefit was gone by day 7, consistent with the drug's short half-life. In the MMPOWER-3 Phase 3 trial (N=218) in primary mitochondrial myopathy, elamipretide 40 mg/day SC for 24 weeks failed both co-primary endpoints — the six-minute walk distance and the fatigue score. Its single FDA approval (Sept 2025, Forzinity) is for the ultra-rare Barth syndrome on a muscle-strength endpoint in about 12 patients — not fatigue, not CFS. There is no registered or published CFS trial; interest is real but anecdotal. Elamipretide works by binding cardiolipin to stabilize cristae and the electron transport chain. It raised the biomarker and still did not relieve fatigue — a Grade D efficacy signal for this condition.

Best for

Understanding why the mitochondrial rationale, even at its strongest, has not translated into a fatigue benefit

Pricing

Prescription (Forzinity, Barth syndrome only) / no CFS pathway

Source: Gioscia-Ryan/Conley et al., PLOS One 2021 (PMC8282018)

Thymosin alpha-1 has the best fit with the immune-dysregulation theory of ME/CFS and is, unusually, a genuinely well-studied peptide drug — just not for this disease. As Zadaxin/thymalfasin it is approved in 30-plus countries for chronic hepatitis B/C and as a cancer immunoadjuvant, with RCT and meta-analytic support for a delayed virological benefit in hepatitis B (OR ~2.67 at 12 months). Mechanistically it enhances dendritic-cell maturation, T-cell differentiation and natural-killer-cell cytotoxicity — directly relevant to the NK deficit documented in ME/CFS, where cytotoxicity runs at roughly half of healthy-control levels. That is a coherent, attractive rationale. But its definitive large RCT — TESTS, in sepsis, N=1,089 — was negative, a reminder that immune plausibility does not guarantee clinical benefit. CFS-specific data are absent at trial quality: the American ME and CFS Society's own Tα1 page is purely mechanistic, citing no ME/CFS trial, no dose and no patient outcomes. Use is confined to clinician practice reports for 'fatigue with frequent infections' and post-viral or long-COVID subtypes, and patient reports are mixed — some describe benefit, others symptom flares. A theoretical caution follows from the mechanism itself: immune activation in a disease that may involve autoimmune or autoantibody features could plausibly worsen some patients. Grade D for this condition until a trial exists.

Best for

The immune-targeted rationale in post-viral fatigue, understood as a hypothesis rather than an evidenced therapy

Pricing

Prescription / compounded (gray zone); varies by pharmacy

Source: Tuthill et al., Ann N Y Acad Sci 2010

Cerebrolysin is the only candidate with a large human-RCT base, but all of it is in stroke, traumatic brain injury and dementia, and even there the efficacy direction is disputed. Dozens of RCTs and meta-analyses exist for acute stroke, TBI, vascular dementia and Alzheimer's, with cognitive-outcome signals such as ADAS-cog improvement in vascular dementia. However, the independent 2023 Cochrane stroke review found no benefit on death or dependence and flagged a possible increase in non-fatal serious adverse events (RR 2.39) — and positive trials are largely manufacturer-associated. The relevance to ME/CFS is purely the 'brain-fog/cognition' extrapolation, since central-nervous-system involvement is part of the disease picture. But that is extrapolation only: no trial in ME/CFS, neurasthenia or idiopathic chronic fatigue as a primary endpoint has been identified. Cerebrolysin is a porcine-derived neurotrophic peptide mixture given parenterally (slow IV infusion or IM), with trial ranges of 10–30 mL/day for 10–28 days; there is no CFS protocol. Its safety profile is generally mild (dizziness, headache, injection-site reactions) but it is contraindicated in epilepsy, carries anaphylaxis risk given its porcine origin, and the Cochrane non-fatal-serious-adverse-event signal warrants caution. There is no US compounding pathway. Big base, wrong indications, contested, never tested in CFS — Grade D for this condition.

Best for

Illustrating that a large RCT base in the wrong diseases still says nothing about CFS

Pricing

No US human-use pathway; not FDA-approved

Source: Ziganshina et al., Cochrane CD007026, 2023 (PMC10565895)

MOTS-c fits the mitochondrial-energy theory conceptually but has the weakest human footing of the four. There is no completed human efficacy trial for any indication. The supporting evidence is robust rodent data — AMPK activation, improved insulin sensitivity, and roughly doubled running capacity in old mice — plus human biomarker associations: endogenous MOTS-c falls with age and is lower in metabolic disease, and exercise raises it around 12-fold in muscle. But association is not therapy. The first human RCT (Phase 2a, in prediabetes/obesity — not CFS) only began recruiting in February 2026 and has no results. CFS-specific data are entirely absent; interest is purely theoretical and community-driven, which is why MOTS-c ranks last despite an appealing 'mitochondrial exercise mimetic' framing. That framing is also conceptually fraught for a population defined by post-exertional malaise. Dosing in the literature is rodent intraperitoneal; there is no validated human dose, and online 'research' dosing (around 5–10 mg SC weekly) is unverified with no clinical basis. Safety is uncharacterized — no controlled human safety data exist, uncontrolled user reports include palpitations, insomnia and injection-site irritation, and the FDA has flagged immunogenicity and impurity risk for unregulated product. On top of all this, MOTS-c is banned by WADA at all times as an AMPK activator (S4.4.1), with no therapeutic-use exemption available. The most speculative of the four — Grade D.

Best for

Recognizing where the evidence is thinnest — rodent biology and biomarkers, not human treatment

Pricing

No approved product; 'research-use-only' material is unapproved/unverified

Source: Reynolds et al., Nat Commun 2021

Rintatolimod is included as context, not as a peptide recommendation, because it defines the bar no peptide has cleared. It is a double-stranded-RNA TLR-3 agonist — a nucleic-acid drug, not a peptide — and it is the only agent ever to run a Phase III CFS trial with a positive result. In severe chronic fatigue syndrome it improved exercise tolerance by 21.3% versus placebo (p=0.047), a modest and contested but genuinely CFS-specific efficacy signal. Crucially, even this remains FDA-unapproved for CFS despite decades of development, which underscores how difficult efficacy is to establish in this condition. Its presence here is deliberate: when people ask which peptide 'works' for ME/CFS, the honest comparison is that the single agent with real Phase III CFS data is not a peptide at all, and even it has not won approval. For completeness, even nutraceuticals are ahead of peptides — a randomized, double-blind trial of coenzyme Q10 plus NADH reported reduced fatigue perception in ME/CFS. Peptides have nothing comparable: no completed CFS trial exists for any of the four candidates above. Rintatolimod is therefore the reference point that keeps the peptide discussion honest, and the reason every peptide in this review is graded D for proven CFS efficacy.

Best for

Anchoring expectations: the only agent with real CFS trial data isn't a peptide, and still isn't approved

Pricing

Investigational for CFS; FDA-unapproved

Source: Strayer et al., PLOS One 2012 (PMC3303772)